# Is chat the wrong shape for AI aided brainstorming? — retained record

> Retained reddit record from reddit:OpenAI.

Canonical: https://highsignal.app/data/records/1e0647ea15ac5b61

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:OpenAI)  published 2026-10-18 20:57 ingested 2026-10-19 02:39

[original source ↗](https://www.reddit.com/r/OpenAI/comments/1x9e7ci/is_chat_the_wrong_shape_for_ai_aided_brainstorming/)

## retained text



```
Something I keep running into: when I ask ChatGPT to brainstorm, I get one long list. It’s fine, but it’s flat. I can’t say “go deeper on #4, ignore the rest” without another wall of text, and after a few rounds the good idea from earlier is buried 30 messages up.

Brainstorming isn’t really a conversation, it’s a tree. You pick a branch, go deeper, back up, try another one. Chat makes you scroll through that tree as one long line.

I’ve been testing a different shape on weekends: each answer is a bubble, you click one to grow it, and everything stays on one map. Each idea also gets a quick score, so you can see at a glance which branches are worth following.

The part I find more interesting: the same thing works as a tool an AI agent can call. With MCP, an agent can ask for “ideas for X” and get back a small, structured set of scored ideas plus a link to the map, instead of writing the brainstorm into the chat itself. So the agent does the work, and you get something you can click around in rather than read.

Curious how others handle this:

* Do you brainstorm in chat, or move to something else (notes, whiteboard, mind map)?
* Would you want your agent to hand back a structured map instead of text, or is that overkill?


```



document fetched 2026-10-19 00:26 · canonical https://reddit.com/r/OpenAI/comments/1x9e7ci/is_chat_the_wrong_shape_for_ai_aided_brainstorming
